Making sense of the polls

Making sense of the polls

United Kingdom Election – 11 days to go

The United Kingdom election will be held in 11 days on the 7 May. You can read my preview from two weeks ago here.

Since that preview was written, little has changed with the overall polling picture. The current UK Polling Report (UKPR) poll average has Labour and the Tories tied on 33%, the United Kingdom Independence Party (UKIP) on 14%, the Liberal Democrats (Lib Dems) on 8% and the Greens on 5%. In Scotland, the Scottish National Party (SNP) continues to dominate; the latest Scottish YouGov poll has the SNP leading Labour by a huge 49-25 margin with 17% for the Tories.

Based on a uniform swing projection applied to Scotland, and a separate projection applied to England and Wales, UKPR currently has 296 of 650 seats for Labour, 259 for the Tories, 58 for the SNP and a few Others, 19 for the Lib Dems and 18 for the Northern Ireland parties. Owing to sophomore surge effects aiding the Tories, I think it is reasonable to give the Tories 20 more seats at Labour’s expense; that gives the Tories a very narrow 279-276 seat lead over Labour. With 326 seats needed for a majority, Labour is still more likely to form government with the SNP’s support.

The overall tied polling picture masks some differences between the polls. YouGov, Populus, Panelbase and TNS currently have 2-3 point leads for Labour. If these polls are right, Labour is likely to win more seats than the Tories and form government with the SNP’s support. These four polls are Internet-based panel polls.

However, ComRes, Ashcroft, ICM, Opinium and Survation currently give the Tories 1-4 point leads over Labour. The first three listed are phone polls, while the last two are also Internet based polls. If these polls are correct, the Tories are likely to win more seats than Labour. While the Tories would still be short of a majority, they could form government with support from the Lib Dems and possibly UKIP or the Northern Ireland Democratic Unionist Party.

Three of the current forecasts have the Tories and Labour effectively tied on seats, while the other two give the Tories a clear seat lead, though still well short of a majority. These forecasts have the SNP winning 48-55 seats, the Lib Dems 24-28 and UKIP 1-4.

The Lib Dems won 23.6% of the Great Britain vote in 2010 (ie, excluding Northern Ireland, which the major UK parties do not contest). Currently, the Lib Dems are polling only 8%, so the current swing is about 15% against the Lib Dems. The uniform swing projection has the Lib Dems winning 19 seats, but I think they will do even worse than this, since there are many 2010 seats where the Lib Dems had less than 15% support; they cannot suffer swings of 15% in those seats without their vote share becoming negative. As a result, the biggest swings against the Lib Dems would be expected in seats where they did well in 2010.

The current forecasts for the Lib Dems to win 24-28 seats appear to rely on Lord Ashcroft constituency polls that show the Lib Dems doing relatively well in seats they hold. However, a ComRes poll of Lib Dem vs Tory seats in southwest England has the Tories leading the Lib Dems by 44-26, a huge turnaround from 2010 when the Lib Dems defeated the Tories by 8.5 points across the 14 seats polled. Ashcroft’s polls had the Lib Dems holding many of these seats.

In my opinion, Ashcroft’s use of a separate constituency voting intention question is flawed, as UK voters have only one vote. I believe that the Lib Dems will perform closer to the Ashcroft standard voting intention question, which had similar results to ComRes. Incumbency will not help them, as it is already factored in; the Lib Dems lost net seats in 2010, so they failed to add many seats that might benefit from a sophomore surge.

The ComRes poll agrees with expectations given the overall voting picture, while the Ashcroft polls disagree. I think that ComRes is closer to the truth here, and that the Lib Dems will be smashed at this election. This will give the major parties more seats than the current forecasts indicate. This would advantage Labour more than the Tories, who need more Lib Dems to form government.

Found this article useful? A tax-deductible gift of $30/month helps deliver knowledge-based, ethical journalism.